| Abstract: | So as to find models suitable for the prediction of water infiltration in purple soil, infiltration rates simulated by three popular empirical infiltration models, Horton model, Kostiakov model, and the modified Kostiakov model, were compared by using those observed results on slope in purple soil under rainfall simulation. 28 sets of infiltration data were divided into two parts for parameter estimation and validation independently. The performance of the three infiltration models was evaluated with the Nash and Sutcliffe efficiency criterion. The results show that the percent for Nash and Sutcliffe efficiency greater than 0.7 is about 69.7% for the Horton model, and 48.48%, and 54.55% for the Kostiakov model and the modified Kostiakov model, respectively. Horton model is better than Kostiakov model and the modified Kostiakov model in most rain events in the field. However, Kostiakov model is still suitable for reasonable prediction of the infiltration in purple soil sloping farmland because fewer parameters are needed. Furthermore, rainfall intensity has significant effects on the parameter values used in the simulation models. The simulation results can be improved by choosing suitable parameters according to the rainfall intensity. |
